The present invention relates to a method for coincidence among common data which is distributed and held in a distribution system having one or more elements including a computation device to enter or withdraw into/from the system.
In a distribution system where plural data bases holding master data and its partial sets as duplicate data are distributed, as to duplicate data update, a data update control method and apparatus for distribution data base system disclosed in Japanese Unexamined Patent Publication No. H10-320356 is known.
In this method, the master data base side holding the master data holds an update history file, and upon master data update, the master data base side-updates the update history file. Upon reception of reference request, the update history file of the master data is referred to, and if the file has been updated, the data of the master data base is obtained. By this arrangement, the master data base and the duplicate data base can be independently updated, and time necessary for the master data base update can be reduced.
In the common-data coincidence method in the conventional distribution system as described above in which master data exists, a device which holds the master data and never withdraws from the system must be provided, and such device usually is a redundant element. Otherwise, if the device holding the master data withdraws from the system, the system cannot function.
Further, there is a high probability that the system cannot function due to the above feature when the device holding the master data is broken. To avoid such inconvenience, plural devices must hold the master data. In this case, such devices may be redundant elements.
Further, an abnormality detection device of multiple processing system to detect valid data from the abovementioned multiplexed data performs detection based on the assumption that the number of devices which output valid data is large. Accordingly, in a system controlled by a human manager or the like, a problem occurs when a large number of outputs are not always valid data. Further, in a case where the number of element devices is too large, as coincidence among all the data cannot be obtained without difficulty, problems occur in the above-described method.